README.md

In epicmodel: Causal Modeling in Epidemiology

epicmodel

Overview

epicmodel is short for “Causal Modeling in Epidemiology” and wants to offer all necessary tools for a causal modeling workflow in R for epidemiologists. Causal modeling describes a structured process of making causal assumptions based on which an epidemiological study is conducted and its results are interpreted. We are always making causal assumptions, at least implicitly. Causal modeling is about doing so explicitly. Did you ever wonder what to measure, how to define your variables, or how to model your outcome of interest? If yes, chances are you need to think about your causal model in more detail.

Causal models are created by making causal assumptions (i.e., that variable A causes variable B) within a causal modeling framework. The current version of epicmodel focuses on one of these frameworks, sufficient-component cause (SCC) models, and offers a way to create them using R. SCC models describe, which sets of causes are in combination sufficient for the outcome of interest to occur.

The package documentation contains many terms with a specific meaning in the context of this package. Check the glossary for an overview: vignette("glossary").

Usage

Creating SCC models follows a three-step workflow (see vignette("epicmodel") for an overview):

1. Create the input for SCC model creation, the so called steplist, using the built-in shiny app. See vignette("steplist") for details.
2. Let epicmodel create the SCC model from the steplist

3. Use the SCC model, e.g., for:

4. Estimating standardized effect size

5. Investigating the effect of prevention and intervention
6. Inspecting the mechanisms behind sufficient causes
7. Transforming the SCC model to a directed acyclic graph (DAG)

Installation

For the latest release:

install.packages("epicmodel")

For the development version:

# install.packages("devtools")
devtools::install_github("forsterepi/epicmodel")

epicmodel documentation built on April 12, 2025, 1:59 a.m.